Phenotypic stratification still has plenty of scope for aiding the tractability of genetic analysis in depression. There are many traits which could be used to create subgroups, including treatment response and physical comorbidities, and perhaps these will be more successful than sex and recurrence. However, it is worth noting that due to the limitations of statistical power with current sample sizes, the performance of the stratified phenotypes presented here is a lower bound of the stratification strategy. Recently, age-at onset9 and use of polygenic risk scores derived from health-related traits73 have been shown to result in subsets of depression with improved heritability. Our study suggests that, until a better understanding of the determinants of genetic heterogeneity in depression exist—increasing sample number remains the optimal strategy.
